The Agent Skills Directory

[PROMPT_INJECTION] (LOW): The skill presents an indirect prompt injection surface where user-provided strings (site name, app name) are used within shell commands during the testing procedure.
Ingestion points: Site name and app name inputs defined in SKILL.md.
Boundary markers: None present in the provided instructions to delimit user input from command logic.
Capability inventory: The skill executes subprocess commands via the bench CLI and writes executable .py and .js files to the local filesystem.
Sanitization: No explicit sanitization or validation of user inputs is mentioned in the markdown procedures.
[COMMAND_EXECUTION] (SAFE): The skill utilizes the bench command-line utility to run tests, migrate sites, and setup environments. These actions are the primary and intended purpose of the skill for developers working with the Frappe framework.
[EXTERNAL_DOWNLOADS] (SAFE): The skill references standard installations of frappe-bench via pip and cypress via npm. These are widely used, reputable tools for the described ecosystem and are fetched from trusted public registries.
[CREDENTIALS_UNSAFE] (SAFE): Documentation examples in references/ci-testing.md and references/cypress.md use default credentials (e.g., 'root', 'admin'). These are clearly provided as placeholders for development and CI/CD service containers rather than being hardcoded production secrets.

frappe-testing